Osteocyte-based image analysis for quantitation of histologically apparent femoral head osteonecrosis: application to an emu model

Femoral head osteonecrosis is often characterized histologically by the presence of empty lacunae in the affected bony regions. The shape, size and location of a necrotic lesion influences prognosis, and can, in principle, be quantified by mapping the distribution of empty lacunae within a femoral head. An algorithm is here described that automatically identifies the locations of osteocyte-filled vs. empty lacunae. The algorithm is applied to necrotic lesions surgically induced in the emu, a large bipedal animal model in which osteonecrosis progresses to collapse, as occurs in humans. The animals' femoral heads were harvested at sacrifice, and hematoxylin and eosin-stained histological preparations of the coronal midsections were digitized and image-analyzed. The algorithm's performance in detecting empty lacunae was validated by comparing its results to corresponding assessments by six trained histologists. The percentage of osteocyte-filled lacunae identified by the algorithm vs. by the human readers was statistically indistinguishable.